1 / 34

The Mazurka Project

The Mazurka Project. CHARM Symposium Kings College London 6 January 2006. Rough score alignment by tapping to recording. For human-assisted alignment of performances to score. Basic evaluation of tapping quality. How accurate is reverse conducting? / Quality metrics.

gauri
Télécharger la présentation

The Mazurka Project

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Mazurka Project CHARM Symposium Kings College London 6 January 2006

  2. Rough score alignment by tapping to recording For human-assisted alignment of performances to score • Basic evaluation of tapping quality How accurate is reverse conducting? / Quality metrics • Raw input to automated alignment process Andrew will present current state of automatic alignment • Website for data and analysis results: http://mazurka.org.uk Recent Progress • Collected about 400 performances of mazurkas on 34 CDs

  3. Sample Performance Listing Duration Performer (year) Label Full list of collected performances at: http://mazurka.org.uk/info/discography

  4. Reverse Conducting and Score Alignment

  5. Reverse conducting of same performance 20 times • used to smooth out errors in individual sessions • used to test various automatic windowing methods (such as SD) • takes about 3 hours / performance to record and process 20 trials • about 400 performances of the mazurkas collected • 20 performances have been reverse conducted thus far: So 2 performances can be tapped per day 7 performances of Op. 7, No. 2 in A Minor 6 performances of Op. 7, No. 3 in F Minor (initial test case) 7 performances of Op. 17, No. 4 in A Minor Lots of Tapping

  6. **kern **beat **abstm **deltatime =1- =1- =1- =1- 4 1 0 0 4 2 391 391 4 3 741 350 =2 =2 =2 =2 4 1 1080 339 4 2 1454 374 4 3 1807 353 =3 =3 =3 =3 4 1 2108 301 4 2 2448 340 4 3 2785 337 =4 =4 =4 =4 4 1 3108 323 4 2 3472 364 4 3 3812 340 Data Entry Method • backbeat.exe: command-line program for recording tap times Example output: • Absolute time from first click in milliseconds. • Other 3 fields for error checking (during performance and afterwards). • Computer keyboard resolution is 5 milliseconds. Beat duration (quarter note) Metric position Absolute Time (ms) Delta Time (ms) Space bar records beats Any letter records barline and first beat of measure barline http://mazurka.org.uk/info/revcond

  7. Offset Alignment with Audio • Need to align first tap to first note in recording. • Cannot just measure start time of note in audio. • Individual tapping trials aligned by least squares fit to a sample of manually measured beat time in audio file. Before alignment After alignment Typical deviations from correct offset: 5, -12, -27, -36 ms Examples of final alignment: Play pid5667230-10-avg (Friedman 1930) Play pid54293-08-avg (Perahia 1994)

  8. Tapping Summary Data **kern **beat **time **dur **min **max **cmin **cmax **sd 4 3 2177 474.95 1935 2265 2147 2206 63.5 =1 =1 =1 =1 =1 =1 =1 =1 =1 4 1 2652 389.1 2536 2832 2624 2679 59.1 4 2 3041 400.15 2921 3137 3017 3065 51.7 4 3 3441 451.65 3399 3493 3429 3453 25.7 =2 =2 =2 =2 =2 =2 =2 =2 =2 4 1 3893 361.85 3845 3957 3879 3906 28.8 4 2 4254 460.4 4206 4283 4245 4264 20.7 4 3 4715 416.3 4620 4780 4697 4733 38.9 =3 =3 =3 =3 =3 =3 =3 =3 =3 4 1 5131 353.65 5039 5225 5108 5154 48.6 4 2 5485 362.45 5433 5541 5471 5499 29.8 4 3 5847 355.45 5780 5911 5830 5864 36.5 =4 =4 =4 =4 =4 =4 =4 =4 =4 4 1 6203 363.7 6160 6266 6189 6216 28.2 4 2 6566 490.25 6490 6602 6552 6580 29.9 4 3 7057 368.2 6980 7154 7036 7077 43.1 =5 =5 =5 =5 =5 =5 =5 =5 =5 4 1 7425 279.65 7318 7478 7407 7443 38.4 4 2 7704 397.85 7657 7760 7693 7716 24.1 4 3 8102 426.65 8058 8160 8089 8115 27.6

  9. **time **kern **kern =1- =1- =1- * *^ * 2465 ([2.C/ 8FF\L 2.r 2659 . 8EEn\J . 2852 . 2CC\ . 3243 . . . =2 =2 =2 =2 3604 4C/] [2.FF\ 2.r 3921 4D-/ . . 4261 4BBn/ . . =3 =3 =3 =3 4569 [2.C/ 8FF\L] 2.r 4759 . 8EEn\J . 4935 . 2CC\ . 5279 . . . =4 =4 =4 =4 5604 4C/] [2.FF\ 2.r 5928 4D-/ . . 6291 4BBn/) . . =5 =5 =5 =5 Score Alignment and Time Interpolation Abs times Score

  10. Output data to Matlab Created from timed score + tapping summary data %%%col01: abstime (average absolute time in milliseconds of human beats) %%%col02: duration (expected duration in ms based on score duration) %%%col03: note (MIDI note number of pitch) %%%col04: metlev (metric level: 1 = downbeat; 0 = beat; -1 = offbeat) %%%col05: measure (measure number in which note occurs) %%%col06: absbeat (absolute beat from starting beat at 0) %%%col07: mintime (minimum absolute time of human beat for this note) %%%col08: maxtime (maximum absolute time of human beat for this note) %%%col09: sd (standard deviation of human beat time in ms.) 2465 1456 48 1 1 0 2419 2535 24.1 2465 194 41 1 1 0 2419 2535 24.1 2659 193 40 -1 1 0.5 -1 -1 -1 2852 752 36 0 1 1 2762 2947 52.4 3604 1155 41 1 2 3 3550 3648 19.8 3921 340 49 0 2 4 3879 3978 26.1 4261 308 47 0 2 5 4239 4275 9.2 4569 1359 48 1 3 6 4548 4585 11.6 4759 176 40 -1 3 6.5 -1 -1 -1 4935 669 36 0 3 7 4906 4968 18.7 5604 1235 41 1 4 9 5585 5628 14.8 5928 363 49 0 4 10 5894 5977 22.2 6291 367 47 0 4 11 6241 6317 15.8 6658 4438 48 1 5 12 6636 6682 13.1 6839 175 40 -1 5 12.5 -1 -1 -1

  11. Tapping Quality Measurements

  12. Manual correction of the beat times • Align tapped beats within 10 ms by ear/eye in sound editor • Each beat alignment takes about 1-2 minutes on average • 300 beats in each mazurka = 1 to 2 days for a performance • Necessary for evaluation of automatic alignment • 4 manual corrections done to date (for 7-2 & 7-3) Play pid5667230-10-corr (Friedman 1930)

  13. Mazurka in F Minor, Op. 7, No. 3 Mazurka in A Minor, Op. 7, No. 2 Rosen 1989 Chiu 1999 Friedman 1930 Friedman 1930 Learning Curves for 4 Performances Play pid9048-06-avg (Chiu 1999)

  14. Average Displacement Errors Mazurka in F Minor, Op. 7, No. 3 Rosen 1989 millseconds = individual trial average displacement errors = dropping more and more later trials = dropping more and more earlier trials

  15. Mazurka in A Minor, Op. 7, No. 2 Chiu 1999 Friedman 1930 Correction Offsets The top plots show amount of time in milliseconds between corrected beat times and average manually tapped beat times. Lower plots display a histogram of same. Spike at 0 in histograms due to 10 ms audible corrections resolution. 60 ms avg correction; -36 ms overall shift 49 ms avg correction; -12 ms overall shift

  16. Mazurka in F Minor, Op. 7, No. 3 Rosen 1989 Friedman 1930 Correction Offsets (2) 48 ms avg correction; -27 ms overall shift 48 ms avg correction; +5 ms overall shift

  17. pid9048-06 (7-2; Chiu 1999) percent A B C E D 0-20 20-40 40-80 80-160 160-320 Milliseconds: Beat Accuracy Metric Logarithmic scale to measure differences between tapped and true beat location: • Human tapper: 48% of beats within 40 milliseconds Pid5667230-10 (7-3; Friedman 1930) horrible bad excellent fair poor good

  18. Automatic Score Alignment

  19. Current Work • Tempo Plots • Tempo Correlation Analysis • Performance Reconstruction

  20. Tempo Plots max 95%avgmax avg Tempo 95%avgmin min Measure number Individual trials (smaller = earlier trial) & (red = earlier; purple=later trial) Friedman 1930 Op. 7, No. 3

  21. Tempo Plots Op. 7, No. 3 (rubato) Mazurka in F minor Op. 7, No. 3 Friedman 1930 Manual corrections: Red dot = beat 1 Blue dots = beats 2 & 3 Play pid5667230-10-03m Rosen 1989 Play pid52932-05-03m

  22. Tempo Plots Op. 7, No. 3 (2) Mazurka in F minor Op. 7, No. 3 (boundaries) Friedman 1930 Play pid5667230-10-10m (Note surprise or lack of it) Rosen 1989 Play pid52932-05-10m

  23. Tempo Plots Op. 7, No. 2 (Third beats red) (phrasing) Mazurka in A minor Op. 7, No. 2 Friedman 1930 Very clear phrase boundaries (peaks every two measures): Play pid5667230-09-02m Chiu 1999 Play pid9048-06-02m

  24. Tempo Plots Op. 7, No. 2 (2) (Third beats red) Mazurka in A minor Op. 7, No. 2 (phrasing) Friedman 1930 Play pid5667230-09-13m Chiu 1999 Play pid9048-06-15m

  25. Tempo Correlation Pearson's product moment correlation: • Correlation value in the range from -1 to +1. • 1 means exact correlation, 0 means no correlation, -1 is anticorrelation • Used in the Krumhansl-Schmuckler key-finding algorithm • Other types of correlation metrics, such as: Spearman Rank Correlation Coefficient

  26. Magaloff 1977 Horowitz 1985 Chiu 1999 Chiu 1999 Magaloff 1977 Horowitz 1985 Chiu 1999 Chiu 1999 Tempo Correlation (2) Op. 17, No. 4 r = 0.75 r = 0.58 Beat durations: r = 0.81 r = 0.57 Measure durations:

  27. Tempo Correlation (3) correlation extremes Raw and corrected reverse conduction: Comparing two unrelated pieces: r = 0.07 r = 0.87 17-4; Chiu 1999 corrected 7-2; Chiu 1999 7-2; Chiu 1999; raw

  28. Tempo Correlation (4) Autocorrelation with shifted performance 7-2; Chiu 1999 measures phrase r-value beat number 17-4; Chiu 1999 r-value beat number

  29. Performance Reconstruction • Simulate performances of the score from various components: Tempo • Constant tempo • Measure tempo • Beat tempo • Tempo of offbeats • Exact duration of all notes • Constant Loudness • Chordal Loudness • Note Loudness boring phrasing jazzing Non-simultaneous beat events Dynamics boring dynamics voicing Also durations for: staccato, legato & pedaling

  30. Abs times Score **time **kern **kern =1- =1- =1- * *^ * 2465 ([2.C/ 8FF\L 2.r 2659 . 8EEn\J . 2852 . 2CC\ . 3243 . . . =2 =2 =2 =2 3604 4C/] [2.FF\ 2.r 3921 4D-/ . . 4261 4BBn/ . . =3 =3 =3 =3 4569 [2.C/ 8FF\L] 2.r 4759 . 8EEn\J . 4935 . 2CC\ . 5279 . . . =4 =4 =4 =4 5604 4C/] [2.FF\ 2.r 5928 4D-/ . . 6291 4BBn/) . . =5 =5 =5 =5 Performance Reconstruction (2) First Reconstruction: • Use tap timings to control the tempo of each beat • Interpolate expected times of offbeats • Convert score to MIDI using **time data with inferred durations. • 7-2; Chiu 1999 Play pid9048-06 • 7-2; Chiu 1999 reconstruction Play pid9048-06-rA • simultaneously Play pid9048-06-sim

  31. Future Work • Audio: • Minimize alignment errors/Speed alignment process • Automatic alignment of offbeats after beats are verified • Non-simultaneous chord note timing offsets • Note dynamics • Performance Analysis: • Characterize and compare performances • Identify importance/relation of timing and dynamics Automatic identification of “schools” of music?

  32. Miscellaneous Slides

  33. clicks: 3 4 5 10 20 JND: 0.0265 0.0115 0.0082 0.0040 0.0024 fastertempo: 63.3 62.1 62.0 62.2 62.8 slower tempo: 56.9 57.0 58.0 57.9 57.3 Tempo Perception Experiment 10 clicks; 60 MM JND * 10 JND * 4 JND JND/4 JND/10

  34. Mazurka in F Minor, Op. 7, No. 3 Mazurka in A Minor, Op. 7, No. 2 Rosen 1989 Chiu 1999 Friedman 1930 Friedman 1930 Average Displacement Errors (2) slower tempo

More Related